Big data junkies still the biggest losers, even though / because they admitted they've been wrong all along

Copying a comment and my replies into a new post, now that the topic of big data and predictions are coming back into the air, as Trump has effectively secured the nomination and likely faces Hillary in the general.

- - - - -

TGGP said:

Since I was predicting the opposite, it behooves me to say that you were right and I was wrong. I really underestimated how much support Trump could get and how much antipathy there was for Cruz in the Republican party. Nevertheless, I still predict he'll lose to Hillary in November, and those who disagree are welcome to bet against me and take my money, as could have occurred earlier.

My response:

Cheer up. The Trump army is going to Make America Great Again, and naysayers casting their hexes in prediction markets are not going to have any more effect than the tens of millions of dollars wasted in attack ads.

In the real world (outside intro stats textbooks), successes and failures are correlated. What failed for the GOP Establishment will fail for the Dem Establishment. They are effectively the same opposition, so one's attack will be a barely mutated version of the other's. And those who got it backward in the primary will get it backward in the general.

As bad as Bush was in the primary, Crooked Hillary will be as bad or worse in the general.

More interesting bets would be -- in which states does she survive? In which ones do write-in Bernie votes outnumber Hillary votes? How big is Trump's margin of victory in Michigan? Etc.

(Again, I'm not interested in robbing people blind over the internet. Just pointing out more interesting ways for the BIG DATA people to spend their time for the next six months.)

To re-iterate a point from this earlier post on "big data being the biggest loser" in this election:

Neither one of us has enough money to make the bet interesting. I would basically require you to risk $1 million, $10 million, etc. Something that would really stop and make you think.

Your idea of "skin in the game" is accepting daunting odds, rather than the absolute magnitude you would lose if wrong.

Not just your idea -- that's the whole approach of prediction markets, which show something like a probability bounded between 0 and 1. No mention of how much the losers stand to lose if they're wrong -- $1, $100, $10 million?

If we look at people who do have a shitload of money to lay on the line, and who normally do so in primary battles -- notice how many of them sat it out. Sheldon Adelson and the Koch brothers stayed out entirely, and most of the early big donors declined to put any more at risk when it was just down to Cruz and Trump.

Mega-donors staying out of the betting pool is a far more honest signal of how uncertain the outcome was, than whatever number of hundred-dollar or even thousand-dollar bets were being placed on the internet.

I still stick by the statement that it was highly uncertain what would happen. If the process played out fairly, it was guaranteed that Trump would win -- that was clear from last fall.

But what was uncertain was the lengths that any number of actors would go to in order to stop Trump from getting the nomination or the Presidency.

That uncertainty is still with us, of course (assassins, etc.), although much less so than before (RNC has capitulated). That's why some of the mega-donors are willing to help Trump's general campaign now.